What actually changed, and how we know it held up

Furthest behind, moving fastest

An earlier read of this data showed a large shift among Arabic-speaking students. A cell-by-cell check caught the cause: an early coding pass had scored their end-of-year answers as a near-uniform "5" on most items, not their actual responses. Corrected, that finding didn't survive. What's below is what did.

Students who started furthest behind gained the most. The effect holds on all six measures tracked for the population with a valid comparison group, from p=.001 to p=.032, and survives every check run against it, including controlling for which school a student was in. In schools where teachers grew stronger as mentors over the year, students grew stronger too (rho≈0.63), strongest specifically on how supported students felt (rho≈0.70). The one place every instrument agrees: self-awareness and reflection are the weak link. Students' own ratings of it fell. Their answers to open scenarios showed thin planning from the start. Teachers turned a reflective eye on their own teaching in only 3% of responses. Classroom observation found no built-in reflection for four in ten students. Four separate measurements, one shared gap. Exactly what the programme should work on next.

Almost 2,000 seventh-graders across 28 classes and four different ed-tech platforms, one school year. The matched analysis behind these numbers covers 704 students and 65 teachers across 21 schools, tracked from the start of the year to the end.

The self-report had it backwards

On self-report scales, Arabic-speaking and Bedouin students rated themselves markedly higher than Hebrew-speaking students on resilience and future orientation. On the one task that's hard to answer flatteringly (where a student picks their own puzzle difficulty instead of rating themselves), that gap nearly disappeared, and reversed slightly. Most of the population difference is how each group uses a rating scale, not a real gap in resilience or mindset. It's exactly the kind of finding a self-report-only evaluation would have missed.

Involvement beat exposure

663 students at the end of the programme, measured against a comparison group in Jewish-society schools. The clearest gap: how concretely students could picture their own future (4.20 among high-schoolers and 4.15 among middle-schoolers, against 3.39 in the comparison group). Involvement beat exposure: being in an actual hi-tech activity, like a tech club, moved a student's intent to take advanced maths roughly three times as much as simply knowing someone in the field (d=0.45 vs d=0.16). One more finding worth naming plainly: girls outscored boys on almost every measure, yet in high school, 15% of girls ruled out advanced maths outright, twice the rate for boys. Whatever's driving that sits outside the mindset measures this instrument tracks. This is a single end-of-programme snapshot, not a before-and-after, so it describes where these students stood, not how far the programme moved them.

Motivation moved. Execution didn't.

Measured over one term, teachers left significantly more motivated to lead innovation in their own teaching and sharper at spotting real problems worth solving (p=.007 and p=.001). Their ability to actually analyse an opportunity and carry an idea through stayed flat (p=.613 and p=.112). The programme built the spark, not yet the follow-through. Precisely where its next version should focus.